Slit Tracing¶

One of the first and most crucial steps of the pipeline is to auto-magically identify the slits (or orders) on a given detector. This is a challenging task owing to the wide variety in:

the number of slits/orders,

the separation between slits/orders (if any)

the varying brightness of flats across the detector

Developing a single algorithm to handle all of these edge cases (pun intended) is challenging if not impossible. Therefore, there are a number of user-input parameters that one may need to consider when running PypeIt (see below).

Underlying the effort is the TraceSlits class which can be used to load the Master frame output for tracing (a FITS and a JSON file).

Algorithm¶

Here is the flow of the algorithms.

A Sobolev S/N image is generated from the trace flat image
edge detection: An initial set of edges are derived from the Sobolev according to the trace-slit-threshold.
match edges: An algorithm is performed to match edges into slits for the first time.
trace (crudely) the slits: Each slit edge is traced with the trace_crude algorithm and snippets of edges are (hopefully) merged
PCA: A PCA analysis is performed of the well-traced edges found thus far. This is then used to rectify the Sobolev images and search for additional edges.
synchronize: Slit edges are synchronized primarily to pick up missing edges
trim: Trimming of small slits is performed

Open Issues¶

Bad columns yield fake edges. These should be masked out by the pipeline using the instrument-specific bad pixel mask.
Overlapping slits are notoriously difficult to detect. One may need to add/subtract individual slits on occasion.

Reduction Mode¶

Longslit¶

If you have only one slit per detector, you may wish to specify the Number of Slits as 1.

Multislit¶

Deriving all of the slits in a mask exposure is challenged by overlapping slits, slits that run to the detector edge, bad columns, etc. Our testing with DEIMOS and LRIS masks is thus far recovering ~95% of the slits.

It is highly recommended that you inspect the warning messages during slit tracing and then pause the code to inspect the MasterTrace output using the pypeit_chk_edges script.

We now summarize the PypeIt parameters that are occasionally varied to improve slit tracing.

One parameter to consider first is the trace-slit-threshold which sets the minimum S/N ratio in the Sobolev filter image for an edge to be detected. You may inspect these edges with the pypeit_chk_edges and –show=edgearr. The left edges in the Sobolev are the white regions in this image and the black regions (negative values) are the right edges. The green/red traces show the left/right edges detected from this image; these are not the final traces. Inspect the positive/negative values of the edges in the Sobolev image and lower/raise trace-slit-threshold accordingly.

If your spectra span only a modest fraction (~50%) of the detector in the spectral direction, you may need to: (1) Reduce the value of trace-slit-mask_frac_thresh and maybe also: (2) Modify the range for smashing the Sobolev image with trace-slit-smash_range.

Add User Slits¶

The code may be instructed to add slits at user-input locations. The syntax is is a list of lists, with each sub-list having syntax (all integers): det:y_spec:x_spat0:x_spat1 For example:

[calibrations]
  [[slits]]
    add_slits = 2:2000:2121:2322,3:2000:1201:1500

The above will add one slit on detector 2 with left/right edge at 2121/2322 at row 2000. The shapes of the slit will be taken from the ones that are nearest.

See the PypeIt-HOWTO-slits slides for further details.

Remove Slits¶

The code may be instructed to remove slits at user-input locations. The syntax is a list of lists, with each sub-list having syntax (all integers): det:y_spec:x_spat For example:

[calibrations]
  [[slits]]
    rm_slits = 2:2000:2121,3:2000:1500

This will remove any slit on det=2 that contains x_spat=2121 at y_spec=2000 and similarly for the slit on det=3.

See the PypeIt-HOWTO-slits slides for further details.

Echelle¶

The primary difference currently between multi-slit and echelle is that the latter analyzes the left and right edges separately during the PCA algorithm.

Scripts¶

pypeit_chk_edges¶

PypeIt includes a simple script to show the processed Trace image and the slit/order edges defined by the algorithm. These are displayed in a Ginga viewer. Here is an example call:

pypeit_chk_edges MF_keck_lris_blue/MasterTrace_A_1_01

If debugging poor performance, you can show other outputs from intermediate steps in the process with the –show command:

--show=edgeearr  # Shows the edges derived early on from the Sobolev image
--show=xset      # Shows the edges derived after the mslit_tcrude() method
--show=siglev    # Shows the Sobolev S/N image

Trace Slit Settings¶

The following are settings that the user may consider varying to improve the slit tracing.

Number of Slits¶

Ironically, one of the more challenging slit configurations to automatically identify is a single slit. In part this is often because at least one edge of the slit butts against the detecor giving no image gradient. And also because only a small portion of the detector may be illuminated by this ‘long’ slit.

Therefore, when reducing long slit data, it may be a good idea to explicitly tell PypeIt that there is only 1 slit to be identified. You can set this using the keyword:

[calibrations]
  [[slits]]
    number=1

You can also use this variable to specify the number of slits that should be detected. Note, that this feature works best when you have well-defined and uniformly illuminated slits (usually the case with cross-dispersed data, for example).

Detection Threshold¶

The detection threshold for identifying slits is set relatively low to err on finding more than fewer slit edges. The algorithm can be fooled by scattered light and detector defects. One can increase the threshold with the sigdetect parameter:

[calibrations]
  [[slits]]
    sigdetect = 30.

Then monitor the number of slits detected by the algorithm.

Presently, we recommend that you err on the conservative side regarding thresholds, i.e. higher values of sigdetect, unless you have especially faint trace flat frames.

On the flip side, if slit defects (common) are being mistaken as slit edges then increase sigdetect and hope for the best.

Fraction Threshold¶

In an interemediate step, the mslit_tcrude() method, the edges defined thus far are traced across the detector with the trace_crude method. A PCA analysis of these is then performed on those edges which spanned at least mask_frac_thresh of the detector in the spectral direction. The default value is 0.6 which may be too large for some instruments (e.g. LRISb with the 300 grism). Consider lowering the value, especially if the code raised a warning on too few edges for the PCA:

[calibrations]
  [[slits]]
    mask_frac_thresh = 0.45

You may also need to adjust the trace-slit-smash_range parameter.

Smash Range¶

One of the final steps in slit/order definition is to identify edges by smashing a rectified version of the Sobolev image. The default is to smash the entire image, but if the spectra are primariliy in a subset of the image one should consider modifying the default parameter. This is frequently the case for low-dispersion data, e.g. LRISb 300 grism spectra (which has a different default value). Modify it as such:

[calibrations]
  [[slits]]
    smash_range = 0.5,1.

Slit Profile¶

DEPRECATED

With relatively short slits (often the case with multiobject or echelle data), the sky background is determined from relatively few pixels towards the edge of the slit, where the flux from a uniformly illuminated slit tends to roll off. To correct for this effect, PypeIt models the spatial slit profile of a trace frame (i.e. a flatfield with the same slit length as the science slit). The relevant set of parameters that determine the fit properties are given by:

reduce slitprofile perform False
reduce flatfield method bspline
reduce flatfield params [n]

where n in the last line should be an integer or floating point number.

The default setting is to not calculate the slit profile. To turn on this functionality, the argument of the first line above can be set to True. If the calculation is performed, the second line sets the method that should be used to determine the spatial slit profile.

At this stage, PypeIt only supports the value ‘bspline’, where the knot spacing is set by the third line above. If the argument of reduce flatfield params is n >= 1, PypeIt will place a knot at every n pixels. Otherwise, if n < 1, PypeIt will place a knot at every k pixels, where k=n*N and N is the total number of pixels in the spectral direction. The number of knots in the spatial direction is set automatically by PypeIt, to be twice the number of pixels along the slit. Thus, the user only has the ability to change the number of knots in the spectral direction (i.e. the blaze function). If the spatial slit profile is not calculated, the blaze function will still be calculated using the ‘reduce flatfield’ settings listed above.

Tips on Trace Flat Frames¶

The slit edges are traced using a “trace” frame. If neighboring slits are very close together, you can use a “pinhole” frame to trace the slit centroid.

In the current version of PypeIt, pinhole frames are only used for echelle data reduction. Pinhole frames are usually an exposure of a quartz lamp through a very short (pinhole) slit. Thus, neighboring slit edges of a pinhole frame should be well separated.

Trace frames, on the other hand, usually have the same slit length as the science frame. In cases where neighboring slits are very close together, it is necessary to first define the slit centroid using a pinhole frame, and the slit edges are defined using a trace frame by “expanding” the slits, by giving the following keyword argument:

trace slits expand True

This has been developed for the APF primarily.

For Developers¶

One of the ways the edge-finding algorithm is fooled is via chip defects, e.g. bad columns. It is therefore valuable to mask any such known features with the bad pixel mask when one introduces a new instrument (or detector).

Table of Contents